Nordigen
Approve all creditworthy customers
Untitled.006.jpeg

Insights / Credit Scoring

Boost your credit scoring with Nordigen’s predictive analytics.

 
 

Insights / Credit Scoring

Boost your credit scoring with Nordigen’s predictive analytics

Harness the power of Open Banking with Nordigen’s Credit Scoring product suite. As used by the most innovative Data Science teams in global banks, lenders, brands and fintechs.

Group 62.png
 
 

Trusted by innovators

_intro.png
 

Credit Scores

 

Why use them?

How it works

Credit scoring models are built, using features that are constructed only from the transaction data. A set of features is passed as a data structure to a model where a score is calculated, based on each of the features and their weight in the model. Thus the output of credit scores is the calculated probability of a customer to default on a loan, based on patterns identified in the transaction data.

Supported across all countries where account data aggregation is available.

  • Credit scores are highly predictive and can be used as both standalone scores and a supplementary input for internal scoring models;

  • Credit scores can be tailored both for optimising approval rates and default rates;

  • Credit scores built from transaction data are proven to be stable over time.

 

Example

 

Credit score output is the calculated probability of a customer to default on a loan, based on patterns identified in the transaction data.

NB: Value 0.7 means that there is a 70% chance that the customer is going to default on a loan.

{“scores”: [{“id”: “82sig7afdg7sf8g”“date”: “2016-04-06”“value”: 0.7}]}Output example:
 

Modelling Features

 

How it works

Features are representations from transaction data that are used as an input for scoring models. Features vary from simple representations (e.g. the average salary) to complex ones (e.g. behaviours that explains why the customer has taken a loan).


Supported across all countries where account data aggregation is available.

Why use them?

  • More than 2,000 predictive features available;

  • All features are standardised across all geographies and can be easily applied.

 

What our customers say

 
Nordigen variables related to loans and financial customer services are consistently found to be top predictors according to the feature importance analysis.
— - Mykola Herasymovych, Creditstar Principal Data Scientist
 
 

Example

 

Features vary from simple representations (e.g. the average salary) to complex ones (e.g. behaviours that explains why the customer has taken a loan) and can be used for improving scoring model accuracy.

{“features”: {“salary_average_month”: 1000,“predicted_next_6_month_average_salary”: 1200,“predicted_next_1_month_average_gambling_spendings”: 200,“nightlife_count_avg_month_12_month_period”: 20,“personal_transfers_max_12_month_period”: 600}}Output example:
 

Decision Rules

 

How it works

A decision rule is a feature with an already applied threshold that determines the group the client belongs to, based on their probability to default.

Supported across all countries where account data aggregation is available.

Why use them?

  • Decision rules are convenient for identifying specific customer groups (e.g. very low-risk customers);

  • Easy to apply in the decision-making process;

  • Decision rules are already optimized feature application for specific geographies and lending products.

 

Example

 

The output is a list of features, their threshold values and an indicator whether the feature value should be above or below the threshold value.

{“decision_rules”: {“average_salary_a_month”: {“value”: 1000,“threshold”: 1200,“is_over_treshold”: False}}}Output example:
 

How can we help you grow?